Pornpinun Chantapacdepong - PowerPoint PPT Presentation

1 / 19
About This Presentation
Title:

Pornpinun Chantapacdepong

Description:

This model is developed by Baillie, Bollerslev and Mikkelsen (1996). 10 ... stable GARCH model as in Baillie and Bollerslev (1990) and the ARCH-in-mean ... – PowerPoint PPT presentation

Number of Views:67
Avg rating:3.0/5.0
Slides: 20
Provided by: Brow186
Category:

less

Transcript and Presenter's Notes

Title: Pornpinun Chantapacdepong


1
Pornpinun Chantapacdepong
  • Fractional Integration and the forward premium
    puzzle.
  • Thammasat University, 6 May 2008.

2
1. Concepts and motivation
  • The forward premium puzzle is the violation of
    the uncovered interest parity (UIP)
  • The UIP parity states that
  • ?st,tk a b(ft,k - st) et,tk
  • ?st,tk is expected changes in exchange rates.
  • (ft,k - st) is the forward premium
  • H0 a0, b1

3
2. Previous literature on UIP
  • Empirical work finds that b?1 and is usually
    negative e.g. Frankel, 1980 Fama, 1984 Bekaert
    and Hodrick, 1993 (these works use a variety of
    exchange rates against the US dollar).
  • Froot and Thaler (1990) find that the average
    coefficient across some 75 published estimates of
    b is -0.88. In these, a few are positive, but
    none of them have the coefficient, b
    statistically greater than or equal to unity.

4
3. Objectives
  • The objectives of this paper are to
  • Test the efficiency of forward markets for
    foreign exchange in the 10 most commonly traded
    currencies using daily data from 1994 to 2007.
  • Attempts to explain the forward premium anomaly
    by
  • Statistical artefacts of the data
  • Foreign exchange risk premium

5
Statistical artefacts of the data
  • This view pays attention to the time series
    properties of the return on spot exchange rate
    and the forward premium.
  • The explanation is based on
  • 1) The long memory behaviour of the forward
    premium
  • (while the returns on the spot exchange rate are
    stationary)
  • 2) The existence of structural breaks in the
    forward premium

6
Foreign exchange risk premium
  • The foreign exchange risk premium explanation
    states that the forward rates are biased
    predictors of actual exchange rate movements
    because there exists a risk premium on one
    country's currency relative to another's.
  • Domowitz and Hakkio (1985) gives an explanation
    of the risk premia through the conditional
    variance of the market forecast error (the error
    in forecasting the spot rate using the forward
    rate).
  • rpt,tk vart (ft,k stk)

7
4. Methodology
  • Unit root test on the expected exchange rate
    depreciation and the forward premium.
  • Correct the structural breaks in the forward
    premium series.
  • Estimate the long memory parameters of the
    forward premium series using the ARFIMA model.
  • Estimate the risk premia by the conditional
    variance of the market forecast error using the
    FIGARCH model.

8
Fractional Integration (ARFIMA model)
  • This model is useful for series that exhibit
    significant autocorrelation ( or long term
    dependence) between observations widely separated
    in time.
  • The ARFIMA(p,d,q) model for yt is written as
  • F(L)(1-L)d (yt-µ)?(L)et
  • where F(L) and ?(L) are the autoregressive
    polynomial and the moving average polynomial in
    the lag operator L
  • d is the fractional differencing parameter.

9
Fractionally Integrated GARCH model
  • We estimate the risk premia by modeling the long
    memory in the conditional variance for the
    forward rate forecast error series.
  • The Fractionally Integrated Generalised
    Autoregressive Conditional Heteroskedasticity
    (FIGARCH) model explains the fractional
    integration (the long memory) behaviour in the
    conditional variance of the series.
  • This model is developed by Baillie, Bollerslev
    and Mikkelsen (1996).

10
5. Results Forward premium
  • Forward premium are non stationary and are in
    fact highly persistent (corresponds to Crowder,
    1994 Evans and Lewis 1995 and Mark, Wu and Hai
    1993).
  • There is fractionally integrated behaviour in the
    forward premium series, 0.5 lt d lt 1 (corresponds
    to Baillie and Bollerslev, 1994 Maynard and
    Phillips, 2001 and Choi and Zivot, 2005).
  • The spot return in all currencies are stationary
    and follow I(0) process. Thus, it is
    inappropriate to apply conventional regression
    analysis to test the hypothesis of the forward
    rate being an unbiased predictor of the future
    spot rate.

11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
Results Forward premium (cont.)
  • The presence of structural breaks in the forward
    premium series is found to increase the
    persistence of the series
  • Even controlling for regime shifts in the forward
    premium series, the fractional integrated
    behaviour still persists.

15
(No Transcript)
16
(No Transcript)
17
Results Risk premia
  • There is strong evidence of long memory in the
    conditional variance of the forecast error.
  • The long memory parameter in the conditional
    variance specification is significantly different
    from zero and falls into the stationary range.
    Thus the propagation of shocks to the variance of
    forward rate forecast error is proved to occur at
    a slow hyperbolic rate of decay.
  • This means that the conditional variance matters
    in determining deviation of the forward rate from
    the expected future spot rate.

18
6. Conclusion
  • This work corroborates earlier findings that a
    principal reason for rejection of the forward
    rate unbiasedness hypothesis is differences in
    persistence between the forward premium series
    and the spot rate return and the existence of the
    time varying exchange rate risk premium.
  • The presence of structural breaks also proves to
    play an important role in explaining the long
    memory of the forward premium data. Allowing for
    structural breaks reduces the persistence of the
    forward premium across all currencies and model
    specifications. Nevertheless, the forward premium
    still follows the (nonstationary) fractionally
    integrated process.

19
6. Conclusion (cont.)
  • In the risk premia estimation part, this work
    suggests a new method in estimating the foreign
    exchange rate risk premium.
  • In estimating the foreign exchange risk premia,
    the FIGARCH model is found to be econometrically
    superior to regular stable GARCH model as in
    Baillie and Bollerslev (1990) and the
    ARCH-in-mean model as used by Domowitz and Hakkio
    (1985).
  • Finding that the risk premia exists proves that
    the Uncovered Interest Parity does not hold.
Write a Comment
User Comments (0)
About PowerShow.com